ORIGINAL ARTICLES ALIMENTARY TRACT. Bravo Catheter-Free ph Monitoring: Normal Values, Concordance, Optimal Diagnostic Thresholds, and Accuracy

Transcription

1 CLINICAL GASTROENTEROLOGY AND HEPATOLOGY 2009;7:60 67 ORIGINAL ARTICLES ALIMENTARY TRACT Bravo Catheter-Free ph Monitoring: Normal Values, Concordance, Optimal Diagnostic Thresholds, and Accuracy SHAHIN AYAZI, JOHN C. LIPHAM, GIUSEPPE PORTALE, CHRISTIAN G. PEYRE, CHRISTOPHER G. STREETS, JESSICA M. LEERS, STEVEN R. DEMEESTER, FARZANEH BANKI, LINDA S. CHAN, JEFFREY A. HAGEN, and TOM R. DEMEESTER Department of Surgery, Keck School of Medicine, University of Southern California, Los Angeles, California Background & Aims: The Bravo ph capsule is a catheter-free intraesophageal ph monitoring system that avoids the discomfort of an indwelling catheter. The objectives of this study were as follows: (1) to obtain normal values for the first and second 24-hour recording periods using a Bravo capsule placed transnasally 5 cm above the upper border of the lower esophageal sphincter determined by manometry and to assess concordance between the 2 periods, (2) to determine the optimal discriminating threshold for identifying patients with gastroesophageal reflux disease (GERD), and (3) to validate this threshold and to identify the recording period with the greatest accuracy. Methods: Normal values for a manometrically positioned, transnasally inserted Bravo capsule were determined in 50 asymptomatic subjects. A test population of 50 subjects (25 asymptomatic, 25 with GERD) then was monitored to determine the best discriminating thresholds. The thresholds for the first, second, and combined (48-hour) recording periods then were validated in a separate group of 115 patients. Results: In asymptomatic subjects, the values measured using a manometrically positioned Bravo ph capsule were similar between the first and second 24-hour periods of recording. The highest level of accuracy with Bravo was observed when an abnormal composite ph score was obtained in the first or second 24-hour period of monitoring. Conclusions: Normal values for esophageal acid exposure were defined for a manometrically positioned, transnasally inserted, Bravo ph capsule. An abnormal composite ph score, obtained in either the first or second 24-hour recording period, was the most accurate method of identifying patients with GERD. In the United States, 7% of the population experiences heartburn daily and 44% experience heartburn at least once a month. 1 Increased esophageal exposure to acidic gastric juice commonly produces heartburn, but the complaint of heartburn is not always a reliable guide to increased esophageal exposure to acidic gastric juice in the esophagus. 2 Similarly, endoscopic esophagitis almost always is associated with increased esophageal exposure to acidic gastric juice, but only about half of the patients with increased esophageal exposure to acidic gastric juice will have esophagitis. 3 These observations make the diagnosis of gastroesophageal reflux disease (GERD) based on symptoms or endoscopic findings problematic. Consequently, the measurement of esophageal exposure to acidic gastric juice has become a key factor in the evaluation of patients suspected of having GERD. The use of a nasoesophageal ph catheter to measure esophageal acid exposure causes nasal, oral, and pharyngeal discomfort; rhinorrhea; and social embarrassment. Consequently, patients limit their activity and become more sedentary when monitored. 4 This can result in less acid reflux and a falsenegative test result. To avoid these problems, a catheter-free radio telemetric system, the Bravo ph capsule (Medtronic Inc., Minneapolis, MN), was introduced into clinical practice on the basis that it would be better tolerated, less limiting to the patient s activity, and allow a longer period of recording. It was assumed that the normal values used for the catheter-based system were applicable to the Bravo ph capsule system. Recent studies have shown that the 2 systems do not record identical values 5,6 and we have noted in normal subjects a difference between the 2 systems in the recorded values for esophageal acid exposure. Consequently, normal values specific to the Bravo system need to be obtained. The objectives of this study were as follows: (1) to obtain normal values for the first and second 24-hour periods for a manometrically positioned, transnasally inserted, Bravo ph capsule in a series of 50 asymptomatic volunteers and to assess concordance between the 2 periods, (2) to determine the optimal discriminating thresholds to distinguish patients with GERD from normal asymptomatic subjects, and (3) to validate this threshold in a separate consecutive group of patients and to identify the recording period with the greatest accuracy. Materials and Methods Technique of Measuring Esophageal Acid Exposure Esophageal manometry was performed to determine the position of the lower esophageal sphincter (LES). The motility catheter consisted of 8 water-perfused channels with lateral openings placed 5 cm apart and oriented radially 45 from each other. The study was performed with the patient in the Abbreviations used in this paper: GERD, gastroesophageal reflux disease; LES, lower esophageal sphincter by the AGA Institute /09/$36.00 doi: /j.cgh

2 January 2009 BRAVO CAPSULE PH MONITORING 61 supine position. With all recording ports in the stomach, the motility catheter was withdrawn in 1-cm increments every 20 seconds. The position of the LES in centimeters from the nostril was recorded. Three characteristics of the LES were assessed: sphincter pressure, overall length, and abdominal length. A defective LES was defined as having a resting pressure of less than 6 mm Hg, an overall length of less than 2 cm, or an abdominal length of less than 1 cm. 7 The motility of the esophageal body was assessed by administering 10 swallows of a 5-mL water bolus, separated by at least 25 seconds. The amplitudes of contraction at each recorded level were averaged and wave progression was assessed by using a commercially available software program (Polygram Net; Medtronic Inc.). 7 Ambulatory ph monitoring was performed using the Bravo ph capsule. The capsule measures mm and contains a battery, radio transmitter, and an antimony ph electrode at its distal end. The ph electrode samples esophageal ph every 6 seconds and data are transmitted to a receiver unit every 12 seconds. Before use, the capsule s ph electrode was calibrated by submersion in a buffer solution of ph 7.01, rinsed with normal saline, and submerged in a second solution of ph The delivery device and the capsule were inserted through an anesthetized nostril into the esophagus. The capsule was advanced into the stomach to confirm gastric acidity and then pulled back into the esophagus so that the ph sensor was positioned 5 cm above the manometrically determined upper border of the LES. A delivery system anchored the capsule to a fold of esophageal mucosa to maintain it at the appropriate level. The subjects were instructed to remain in the upright or sitting position until retiring to bed in the evening, to refrain from eating or drinking between meals, to avoid chewing gum or smoking, and to go about their normal duties at home or work. One meal during the study was standardized and consisted of a hamburger, fries, and milkshake obtained at a fastfood outlet. The other meals were chosen from a list of foods with a ph between 5 and 7. The meals were to be consumed at one sitting and accompanied only by water, milk, coffee, or tea. Carbonated beverages, alcohol, and fruit drinks with a low ph were excluded. Subjects were instructed to lie flat at night if possible with a single pillow and no blocks under the head of the bed. No medications affecting gastrointestinal function were allowed during the monitored period. Acid-suppression therapy with H2-blockers or proton pump inhibitors was discontinued 1 and 2 weeks, respectively, before the ph monitoring. Subjects maintained a diary of events including the beginning and end of their meals, the time when they retired to bed, and the time when they arose in the morning. They returned at the end of the recording period and the data were downloaded from the recording units to a personal computer for analysis using a commercially available software program (Polygram Net). Six components were used to assess esophageal acid exposure: the percentage of time the ph was less than 4 for the total monitored period and for the time span in the upright and supine positions, the number of reflux episodes, the number of reflux episodes greater than 5 minutes in length, and the length of the longest reflux episode in minutes. 8 Previous experience has shown that some of these components can be normal and others abnormal during the same recording period, making the results of the test difficult to interpret. 8 To overcome this difficulty, a composite ph score has been developed based on the mean and standard deviation of each component derived from a group of 50 normal subjects that weighs each component according to the dependability and reliability of the ph measurement for that component The same formula was used to develop composite ph scores for the first and second 24-hour recording periods and the total 48-hour recording period for the Bravo system. Study Population and Methods for Each Objective Objective 1: to obtain normal values for the first and second 24-hour periods of Bravo recording in a series of 50 asymptomatic volunteers and to assess concordance between the 2 recording periods. Fifty asymptomatic subjects were recruited for monitoring using the Bravo ph capsule to determine normal values for esophageal acid exposure. On interview, the subjects did not have a history of heartburn, regurgitation, dysphagia, atypical symptoms of reflux (nausea, hiccups, globus sensation, chest pain, laryngitis, hoarseness, chronic cough, asthma, or recurrent pneumonia), use of antacid or acid-suppression medication, or previous esophageal, gastroduodenal, or biliary tract surgery. All had a normal barium upper-gastrointestinal study and a normal esophageal motility study. The mean, standard deviation, median, and 95th percentile values were calculated for each of the 6 components used to assess esophageal acid exposure. The composite ph score was calculated for the first and second 24-hour recording periods and the combined 48-hour recording period. Concordance between the first and second 24-hour recording periods was assessed using the Bland and Altman 12 approach. Objective 2: to determine the optimal discriminating thresholds to distinguish patients with GERD from normal asymptomatic subjects. A separate group of 25 asymptomatic volunteers were recruited who met the clinical criteria described earlier for the 50 subjects used to determine normal values for acid exposure. An additional group of 25 symptomatic patients were recruited who had a high symptomatic probability of GERD. All 25 of these patients had heartburn with a median 8.1 episodes per Table 1. Criteria for Control Groups Strong clinical evidence of GERD Primary symptoms of heartburn and/or regurgitation Response to PPI therapy 50% Hiatal hernia 2 cm on barium esophagram Esophageal mucosal injury (endoscopic esophagitis, microscopic inflammatory infiltrate, or intestinal metaplasia) Minimal clinical evidence of GERD Primary symptoms other than heartburn and/or regurgitation Response to PPI therapy 50% No hiatal hernia No esophageal mucosal injury Indeterminate clinical evidence of GERD Patients with mixed characteristics who did not fit clearly into either of the earlier-described groups NOTE. Patients were evaluated by symptom questionnaire, barium esophagram, and endoscopy with biopsy. None had a history of previous foregut surgery. PPI, proton pump inhibitor.

3 62 AYAZI ET AL CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 7, No. 1 Table 2. Mean, Standard Deviation, Median, and 95th Percentile Values for the First and Second 24-Hour Recording Periods and the Combined 48-Hour Study With the Bravo ph Capsule in 50 Asymptomatic Subjects First 24 hours (n 50) Second 24 hours (n 48) a Combined 48 hours (n 48) a Mean (SD) Median 95th percentile Mean (SD) Median 95th percentile Mean (SD) Median 95th percentile % Total time ph (2.16) (1.79) (1.61) % Upright time ph (3.14) (2.60) (2.29) % Supine time ph (1.18) (1.29) (1.04) No. of reflux episodes 21.2 (18.6) (20.1) (33.51) No. of reflux episodes 5 min 0.62 (1.21) (1.16) (1.64) 1 5 Longest episode, min 3.79 (4.31) (7.71) (7.02) Composite ph score 6.00 (4.88) (4.58) (4.80) a Two subjects had incomplete data (probe detached after 18 hours in one subject and after 36 hours in the other subject) and were not included in the analysis. week and were responsive to acid-suppression therapy. Twenty had regurgitation. All had radiographic evidence of a hiatal hernia greater than 2 cm. Twenty had a defective LES on manometry. Thirteen had mucosal injury: 8 had grade II to III esophagitis and 5 had Barrett s esophagus of 3 cm or greater. These 2 groups (25 asymptomatic volunteers and 25 symptomatic patients) were used to determine the discriminating thresholds. Logistic regression analysis was used to determine the optimal discriminating threshold for each component and for the composite ph score. The c-statistic (area under the curve), max-rescaled R 2 statistic, and the concordant percentage for each cut-off point that best differentiates between normal subjects and symptomatic patients were calculated. The threshold with the highest c-statistic, R 2, and concordant percentage was considered the optimal discriminating threshold. The analysis was performed for each component and for the composite ph score for the first and second 24-hour recording periods and for the combined 48-hour recording period. Receiver operator characteristic curves also were constructed to determine the optimal discriminating thresholds for each component and for the composite ph score. The Statistical Analysis System (SAS, version 9.2; SAS Institute Inc, Cary, NC) was used. Thresholds were determined for the composite ph score for the first, second, and the combined 48-hour recording period by selecting the cut-off point with the greatest combination of sensitivity and specificity for discriminating patients from normal subjects. This analysis was performed using Prism 4 statistical software (Graphpad, San Diego, CA). Objective 3: to validate the optimal threshold in a separate consecutive series of patients and to identify the recording period with the greatest accuracy. The optimal thresholds for the composite ph scores for the first and second 24-hour and the combined 48-hour recording periods were applied to a separate consecutive series of 115 symptomatic patients evaluated with the Bravo ph capsule. None of the patients had previous foregut surgery. All patients had a barium esophagram and endoscopy with biopsy. Patients symptoms were assessed by a questionnaire and their response to proton pump inhibitor therapy was recorded. The 115 patients were divided into 3 groups: (1) patients with strong clinical evidence of GERD (n 28), (2) patients with minimal evidence of GERD (n 10), and (3) patients who were indeterminate for the presence of GERD (n 77). The criteria used to define these groups are shown in Table 1. The accuracy of the composite ph score was assessed for each of the recording periods by calculating the sensitivity, specificity, and the positive and negative predictive values. The group of patients with strong clinical evidence of GERD was used as a positive control and the group with minimal clinical Table 3. Concordance of Esophageal Acid Exposure Between the First and Second 24-Hour Periods of ph Recording With the Bravo ph Capsule in 50 Asymptomatic Subjects First 24 hours (n 50) Second 24 hours (n 48) a Mean difference b (n 48) a Mean (SD) Mean (SD) (95% CI) % Total time ph (2.16) 1.78 (1.79) 0.05 ( 0.63 to 0.74) % Upright time ph (3.14) 2.54 (2.60) 0.00 ( 0.99 to 0.99) % Supine time ph (1.18) 0.34 (1.29) 0.00 ( 0.38 to 0.38) No. of reflux episodes 21.2 (18.6) 22.3 (20.1) 0.46 ( 5.96 to 5.04) No. of reflux episodes 5 min 0.62 (1.21) 0.75 (1.16) 0.10 ( 0.61 to 0.40) Longest reflux episode, min 3.79 (4.31) 5.87 (7.71) 2.05 ( 4.37 to 0.26) Composite ph score 6.00 (4.88) 6.05 (4.58) 0.03 ( 1.56 to 1.63) CI, confidence interval. a Two subjects had incomplete data (probe detached after 18 hours in 1 subject and after 36 hours in the other subject) and were not included in the analysis. b Difference first 24-hour value minus second 24-hour value.

4 January 2009 BRAVO CAPSULE PH MONITORING 63 evidence of GERD was used as a negative control. The 77 patients with indeterminate evidence for GERD were classified as refluxer or nonrefluxer using the composite ph score to assess the diagnostic yield of prolonged esophageal ph monitoring in the clinical setting when the diagnosis was frequently ambiguous. Ethical Considerations The study protocol and consent forms were approved by the Los Angeles County and University of Southern California Institutional Review Boards. Results Objective 1: to obtain normal values for the first and second 24-hour periods of Bravo recording in a series of 50 asymptomatic volunteers and to assess concordance between the 2 recording periods. Normal values for the 6 individual components and the composite ph score obtained with the Bravo ph capsule are shown in Table 2. In asymptomatic volunteers, there was excellent concordance for all 6 components and for the composite ph score between the first and second 24-hour recording periods (Table 3). Bland Altman 12 plots showing the 95% limits of agreement for the total percentage time the ph was less than 4 Figure 1. Bland Altman plots comparing (A) total time the ph was less than 4 and (B) the composite ph score as measured by the Bravo ph capsule in 50 asymptomatic subjects on 2 consecutive days. The mean difference between day 1 and day 2 is plotted against (A) the total time the ph was less than 4 and (B) the composite ph score for each subject. The dashed lines represent the 95% limits of agreement ( 2 standard deviations). These limits of agreement define an interval within which 95% of differences between measurements are expected to lie. All but one subject fell within the limits of agreement.

5 64 AYAZI ET AL CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 7, No. 1 Table 4. Optimal Threshold for the Six Components and the Composite ph Score of Esophageal Acid Exposure Measured With the Bravo ph Capsule Optimal threshold c R 2 Percentage concordant First 24 hours (n 50) % Total time ph % % Upright time ph % % Supine time ph % No. of reflux episodes % No. of reflux episodes 5 min % Longest reflux episode, min % Composite ph score % Second 24 hours (n 44) % Total time ph % % Upright time ph % % Supine time ph % No. of reflux episodes % No. of reflux episodes 5 min % Longest reflux episode, min % Composite ph score % Combined 48 hours (n 44) % total time ph % % upright time ph % % supine time ph % No. of reflux episodes % No. of reflux episodes 5 min % Longest reflux episode, min % Composite ph score % c, area under the curve statistic; R 2, max-rescaled R 2. and the composite ph score for the first and second 24-hour periods of Bravo recording are shown in Figure 1. Objective 2: to determine the optimal discriminating thresholds to distinguish patients with GERD from normal asymptomatic subjects. Table 4 shows the optimal discriminating threshold for each component and the composite ph score defined by the highest c-statistic and the corresponding R 2 value. The percentage of concordance reflects the accuracy of the classification of normal subjects and the symptomatic patients into the appropriate group. By using these thresholds, the sensitivity, specificity, and accuracy were calculated (Table 5). The accuracy was greater than 90% for each component and for the composite ph score. The optimal threshold for the composite ph score determined by logistic regression was 14 for the first and second 24-hour recording periods, and 16 for the combined 48-hour recording period. There was variability in the composite ph score between the first and second 24-hour recording periods in symptomatic patients. This was less so in normal subjects (Figure 2). Despite the variability seen in symptomatic patients, all were classified correctly as abnormal on each day using the composite ph score of 14 derived by logistic regression. Receiver operating characteristic analysis also was performed to determine the optimal discriminating threshold for the composite ph score. The most accurate thresholds were for the first 24 hours (93rd percentile), 16.2 for the second 24 hours (98th percentile), and for the combined 48 hours of Bravo recording (99th percentile). Although the thresholds differed slightly, the accuracy of classification of normal subjects and symptomatic patients was not improved by using these thresholds compared with those determined by logistic regression analysis. Objective 3: to validate the optimal threshold in a separate consecutive series of patients and to identify the recording period with the greatest accuracy. The optimal thresholds for the composite ph scores determined by logistic regression in the previous section were validated using a separate consecutive series of 115 patients divided into 3 groups based on clinical likelihood of having GERD (Table 1). There were 28 patients with strong clinical evidence of GERD, 77 with indeterminate clinical evidence of GERD, and 10 with minimal evidence of GERD. In the 28 patients with strong clinical evidence of GERD, 19 (68%) had an abnormal composite score during both days of recording. Only 2 of these patients (7%) had a normal composite ph score on both days. The composite ph score was abnormal during the first, second, or both days of monitoring in 26 patients (93%). None of the 10 patients with minimal clinical evidence of GERD had an abnormal composite ph score during any monitoring period. To determine which recording period most accurately classified patients with strong and minimal clinical evidence of GERD, sensitivity, specificity, positive and negative predictive values, and accuracy were calculated using the composite ph scores for the first 24-hour period, the second 24-hour period, either 24-hour period, or the combined 48-hour period. In the validation group of 115 patients, those with a positive composite ph score and strong clinical evidence of GERD were considered to be true positives and those with a negative composite ph score and minimal evidence of GERD were considered to be true negatives. An abnormal composite ph score during either the first or second day of monitoring was most sensitive for detecting abnormal reflux and most accurately classified patients into the appropriate group (Table 6). Similar results were

6 January 2009 BRAVO CAPSULE PH MONITORING 65 Table 5. Diagnostic Accuracy for GERD Using the Optimal Thresholds of Each Component and the Composite Esophageal ph Score Measured With the Bravo ph Capsule Optimal threshold Sensitivity (95% CI) Specificity (95% CI) Accuracy (95% CI) First 24 hours (n 50) % Total time ph (80 100) 96 (80 100) 96 (86 100) % Upright time ph (74 99) 96 (80 100) 94 (83 99) % Supine time ph (59 93) 100 (86 100) 90 (78 97) No. of reflux episodes (74 99) 88 (69 97) 90 (78 97) No. of reflux episodes 5 min 2 88 (69 97) 92 (74 99) 90 (78 97) Longest reflux episode, min (80 100) 96 (80 100) 96 (86 100) Composite ph score (86 100) 96 (80 100) 98 (89 100) Second 24 hours (n 44) % Total time ph (86 100) 100 (84 100) 100 (92 100) % Upright time ph (68 97) 100 (84 100) 93 (82 99) % Supine time ph (72 99) 100 (84 100) 95 (85 99) No. of reflux episodes (79 100) 90 (70 99) 93 (82 99) No. of reflux episodes 5 min 2 96 (79 100) 95 (76 100) 96 (85 99) Longest reflux episode, min (68 97) 100 (84 100) 93 (82 99) Composite ph score (86 100) 100 (84 100) 100 (92 100) Combined 48 hours (n 44) % Total time ph (85 100) 95 (76 100) 98 (88 100) % Upright time ph (85 100) 95 (76 100) 98 (88 100) % Supine time ph (61 95) 100 (84 100) 91 (78 97) No. of reflux episodes (72 99) 90 (70 99) 91 (78 97) No. of reflux episodes 5 min (88 100) 100 (84 100) 100 (92 100) Longest reflux episode, min (72 99) 100 (84 100) 95 (85 99) Composite ph score (88 100) 100 (84 100) 100 (92 100) CI, confidence interval. obtained using the percentage of total time the ph was less than 4 (data not shown). In the validation group of 115 patients, 77 had indeterminate clinical evidence of GERD. These patients were used to assess the diagnostic yield of the prolonged esophageal ph monitoring in the clinical setting when the diagnosis of GERD was ambiguous. Of these 77 patients, 51 (66%) were classified as having the disease if either the first or second 24-hour monitoring period was used. Conversely, if only the first 24-hour period, the second 24-hour period, or the combined 48-hour recording period were used, the diagnostic yield would be 52%, 55%, and 40%, respectively. Discussion Clinical experience has shown a difference in the values obtained for esophageal acid exposure in normal subjects using the Bravo ph capsule and the conventional naso-esophageal ph catheter. 5,6 Consequently, normal values specific to the manometrically placed Bravo ph capsule and optimal discriminating thresholds to differentiate normal from abnormal esophageal acid exposure needs to be defined for accurate use of the test in the clinical setting. At the present time, studies performed with manometrically placed Bravo ph capsules are interpreted using catheter-based normal values that are not applicable to the Bravo system or normal values from Bravo ph capsules placed endoscopically 6 cm above the squamocolumnar junction. 13,14 There are 2 problems with using the normal values obtained by endoscopic placement. First, is the variability in the position of the squamocolumnar junction in relationship to the LES that occurs in normal subjects and patients with GERD. 15 Second, is the 7-fold increase in acid exposure that occurs over the distal 5 cm of the esophagus above the LES. 16 In addition, with endoscopically placed Bravo capsules there are differences in acid exposure between the first and second 24-hour recording periods as a result of the sedation used for the endoscopy. 17,18 It has been reported that with the Bravo ph capsule there are differences in acid exposure between the first and second 24-hour recording periods. 14,19 One would expect that asymptomatic subjects and patients carefully selected for having a high probability of GERD would be normal or abnormal, respectively, during both monitoring periods. We found a low degree of discordance (5% and 0%, respectively) between the first and second 24 hours in a group of asymptomatic subjects and a group of patients with strong evidence of GERD. It is not surprising that when evaluating patients with indeterminate clinical evidence for GERD there may be a higher degree of discordance between the 2 days. We found a 27% discordance between the first and second 24-hour Bravo ph recording periods in our ambiguous patients. In this situation the discordance may be more a function of the stage of GERD than a problem with the variability of the measurement. This discordance rate in patients with indeterminate clinical evidence of GERD prompted us to compare the sensitivity, specificity, and positive and negative predictive values for the first 24-hour period, the second 24-hour period, either 24-hour period, or the combined 48-hour period. This analysis showed that the diagnostic accuracy was highest when either 24-hour period was used. Therefore, when interpreting the ph results in patients who have discordance between the first and second 24-hour recording periods, we recommend that if either the

7 66 AYAZI ET AL CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 7, No. 1 first or second 24-hour period is abnormal, the patient has increased esophageal acid exposure. The question emerges as, what is the variability between a Bravo ph recording obtained using a capsule placed manometrically or endoscopically? Pandolfino et al 14 have reported a feasibility study consisting of 42 normal subjects and Wenner et al 13 reported a series of 50 normal subjects using a Bravo ph capsule placed endoscopically 6 cm above the squamocolumnar junction. In both studies the 95th percentile normal values for the percentage of time the ph was less than 4 for the total and upright time were similar to our normal values based on logistic regression analysis (total time ph 4 5.9% and 3.3% vs 5.8%; upright time ph 4 7.8% and 5.2% vs 6.2%). However, they had a higher value for the supine time with a ph of less than 4 than we observed in our study (6.3% and 4.9% vs 1.6%). Pandolfino 14 indicated that this value was affected by 4 outliers who had a highly reflux-provoking diet before sleep. Even after elimination of these outliers, their value still is higher than ours. These differences highlight the variability between the 2 methods of capsule placement and statistical methodology used to define normal values. This encourages a comparative study using the 2 techniques of placement and a common statistical analysis. Figure 2. Composite ph score for the first and second 24-hour recording periods using the Bravo ph capsule in 25 normal subjects (solid lines) and 25 symptomatic patients (dotted lines). The horizontal line represents the optimal threshold for the composite ph score (14) determined by logistic regression. Table 6. Diagnostic Precision of the Composite ph Score Using the Bravo ph Capsule Abnormal first 24 hours Abnormal second 24 hours Abnormal combined 48 hours Abnormal either first or second 24 hours In summary, the difference between normal values obtained with manometrically placed Bravo ph capsules and endoscopically placed Bravo ph capsules necessitates the use of normal values specific to the method of placement of the Bravo capsule. In normal subjects and in patients with a high probability of GERD, there was excellent concordance between the first and second 24-hour Bravo recording period when the capsule was placed based on manometric measurements. However, when the test was applied to unselected patients with indeterminate clinical evidence of GERD, there was some degree of discordance or day-to-day variability. Consequently, many patients may be misclassified if only one 24-hour recording period is used. The highest diagnostic accuracy with manometrically placed Bravo ph capsule occurs when the composite ph score is abnormal on either the first or second 24-hour recording period. References Sensitivity Specificity PPV NPV Accuracy 86% 100% 100% 71% 89% 79% 100% 100% 63% 84% 82% 100% 100% 67% 87% 93% 100% 100% 83% 95% NOTE. The group of patients with strong clinical evidence of GERD (n 28) was used as a positive control, and the group with minimal clinical evidence of GERD (n 10) was used as a negative control. PPV, positive predictive value; NPV, negative predictive value. 1. Talley NJ, Zinsmeister AR, Schleck CD, et al. Dyspepsia and dyspepsia subgroups: a population-based study. Gastroenterology 1992;102: Tefera L, Fein M, Ritter MP, et al. Can the combination of symptoms and endoscopy confirm the presence of gastroesophageal reflux disease? Am Surg 1997;63: DeMeester TR, Peters JH, Bremner CG, et al. Biology of gastroesophageal reflux disease: pathophysiology relating to medical and surgical treatment. Ann Rev Med 1999;50: Fass R, Hell R, Sampliner RE, et al. Effect of ambulatory 24-hour esophageal ph monitoring on reflux-provoking activities. Dig Dis Sci 1999;44: Pandolfino JE, Schreiner MA, Lee TJ, et al. Comparison of the Bravo wireless and Digitrapper catheter-based ph monitoring systems for measuring esophageal acid exposure. Am J Gastroenterol 2005;100: Pandolfino JE, Zhang Q, Schreiner MA, et al. Acid reflux event detection using the Bravo wireless versus the Slimline catheter ph systems: why are the numbers so different? Gut 2005;54: Zaninotto G, DeMeester TR, Schwizer W, et al. The lower esophageal sphincter in health and disease. Am J Surg 1988;155: Johnson LF, Demeester TR. Twenty-four-hour ph monitoring of the distal esophagus. A quantitative measure of gastroesophageal reflux. Am J Gastroenterol 1974;62: Streets CG, DeMeester TR. Ambulatory 24-hour esophageal ph monitoring: why, when, and what to do. J Clin Gastroenterol 2003;37:14 22.

8 January 2009 BRAVO CAPSULE PH MONITORING Johnson LF, DeMeester TR. Development of the 24-hour intraesophageal ph monitoring composite scoring system. J Clin Gastroenterol 1986;8(Suppl 1): Jamieson JR, Stein HJ, DeMeester TR, et al. Ambulatory 24-h esophageal ph monitoring: normal values, optimal thresholds, specificity, sensitivity, and reproducibility. Am J Gastroenterol 1992;87: Bland JM, Altman DG. Statistical methods for assessing agreement between two methods of clinical measurement. Lancet 1986;1: Wenner J, Johnsson F, Johansson J, et al. Wireless oesophageal ph monitoring: feasibility, safety and normal values in healthy subjects. Scand J Gastroenterol 2005;40: Pandolfino JE, Richter JE, Ours T, et al. Ambulatory esophageal ph monitoring using a wireless system. Am J Gastroenterol 2003;98: Csendes A, Maluenda F, Braghetto I, et al. Location of the lower oesophageal sphincter and the squamous columnar mucosal junction in 109 healthy controls and 778 patients with different degrees of endoscopic oesophagitis. Gut 1993; 34: Mekapati J, Knight LC, Maurer AH, et al. Transsphincteric ph profile at the gastroesophageal junction. Clin Gastroenterol Hepatol 2008;6: Bhat YM, McGrath KM, Bielefeldt K. Wireless esophageal ph monitoring: new technique means new questions. J Clin Gastroenterol 2006;40: Ahlawat SK, Novak DJ, Williams DC, et al. Day-to-day variability in acid reflux patterns using the BRAVO ph monitoring system. J Clin Gastroenterol 2006;40: Tseng D, Rizvi AZ, Fennerty MB, et al. Forty-eight-hour ph monitoring increases sensitivity in detecting abnormal esophageal acid exposure. J Gastrointest Surg 2005;9: Address requests for reprints to: Tom R. DeMeester, MD, Department of Surgery, Keck School of Medicine, University of Southern California, 1510 San Pablo Street, Los Angeles, California demeester@surgery.usc.edu; fax: (323) The authors disclose no conflicts.